View Issue Details Jump to Notes ] Print ]
IDProjectCategoryView StatusDate SubmittedLast Update
0014615ParaView(No Category)public2014-03-03 15:462014-11-26 20:56
ReporterAlan Scott 
Assigned ToDavid C. Lonie 
PriorityurgentSeverityminorReproducibilityhave not tried
StatusclosedResolutionfixed 
PlatformOSOS Version
Product Versiongit-master 
Target Version4.2Fixed in Version4.2 
Summary0014615: Volume rendering crashes
DescriptionVolume rendering, remote server, now crashes. It may take a bit of work to replicate (Utkarsh couldn't replicate). Here is what I did:

ParaView master. Linux. Remote server. 8 cores. Mesa (software rendered). Edit/ Settings/ Server/ Remote Render Threshold == 0MB. This is a release build.

Open can.exo. All variables on. Apply.
Forward one timestep.
Paint by EQPS.
Change Surface representation to Volume.

Output messages are as follows:

8 of the following:
ERROR: In .../src/ParaView/4.2.0-alpha/ParaView4/VTK/Rendering/VolumeOpenGL/vtkOpenGLProjectedTetrahedraMapper.cxx, line 183
vtkOpenGLProjectedTetrahedraMapper (0x187436b0): The required extensions are not supported.


8 of the following:
ERROR: In .../src/ParaView/4.2.0-alpha/ParaView4/VTK/Rendering/VolumeOpenGL/vtkOpenGLProjectedTetrahedraMapper.cxx, line 268
vtkOpenGLProjectedTetrahedraMapper (0x187436b0): failed at glRenderBufferStorage color 1 OpenGL errors detected
  0 : (1280) Invalid enum


8 of the following:
Warning: In .../src/ParaView/4.2.0-alpha/ParaView4/VTK/Rendering/VolumeOpenGL/vtkOpenGLProjectedTetrahedraMapper.cxx, line 309
vtkOpenGLProjectedTetrahedraMapper (0x18023070): Missing FBO support. The algorithm may produce visual artifacts.




We now get an intermittent crash.

Ask me if you want me to get this in a debugger, or try to get a stack trace.

TagsNo tags attached.
ProjectSandia
Topic Name14615-fix-mpi-segfault
Typecrash
Attached Files

 Relationships
related to 0014581closedDavid C. Lonie Volume rendering is bad remote rendering 

  Notes
(0032448)
Utkarsh Ayachit (administrator)
2014-03-03 17:23

Alan,

Which mesa version are you using?
(0032449)
Alan Scott (manager)
2014-03-03 17:26

Sorry, should have included that. 7.10.3.

Alan
(0032451)
Utkarsh Ayachit (administrator)
2014-03-03 17:28

David,

You'll have to build with Mesa. I believe the problem is stemming from Burlen's changes to vtkOpenGLProjectedTetrahedraMapper.
(0032457)
David C. Lonie (developer)
2014-03-05 14:39

I can reproduce this, I'll start digging around for the cause.
(0032459)
David C. Lonie (developer)
2014-03-06 15:46

The warnings, at least, seem to be caused by Mesa 7.10.3 not supporting GL_ARB_texture_float. I've not been able to reproduce the crash, but updating the logic in the mapper may solve this for you.

Alan, can you test out the VTK patch http://review.source.kitware.com/#/t/4034/ [^] and see if this fixes the crash? It should at least fix the warnings (except for 'Missing FBO support.' -- that one is legit).

Dave
(0032493)
Utkarsh Ayachit (administrator)
2014-03-20 08:32

Topics merged in master
(VTK) 14258-remote-polyhedra
        14580-top-level-checkbox
        14584-frequency-live-view
(VTK) 14614-report-numpy
(VTK) 14615-volume-rendering-tet-crash
        animation-movie-compressor-support
(VTK) catalyst-web-interactive-composite
        data_template_redefine
(VTK) enable_subclassing_vtkControlPointsItems
        ensemble-reader
        expand_fragment_integration
        fix-variable-referece
        fix_qt5_dashboard_failures
(VTK) probe_enhancements
        sciberquest-test-timeout
(0032508)
Alan Scott (manager)
2014-03-20 21:18

Nope, still crashes. Details:

* With regards to the warning, we get a warning for every process. If I have 4000 processes, that is an unacceptable number of warnings. Please cut it back to 1.

* "Missing FBO support. The algorithm may produce visual artifacts" isn't descriptive. Is it possible to elaborate? What type of FBO support? Be more verbose, but do it only if the build is debug. (Actually, it could be that way, I am not sure.)

* Why are we missing FBO support now, but 4.1.0 worked fine without complaint? PV 4.2.0 must continue to do volume rendering, in parallel, remote server, with Mesa (for instance, 7.*.*)

Last, it still crashes. Did you try replicating on 8 servers? Here is the crash dump (I have a debug build, but the output looks like release..):

(I have cut out 7 of every 8 lines of output, for brevity.)

I will send directly to David and Utkarsh

(0032509)
Alan Scott (manager)
2014-03-20 21:20

I forgot - to get to the crash, you need to click in the 3d window.
(0032521)
David C. Lonie (developer)
2014-03-21 12:41

I've managed to reproduce the crash -- it appears a double click triggers it, which triggers a SelectCells operation on the server for the clicked point.

While rendering the geometry to prepare for the pick, the vtkOpenGLProjectedTetrahedraMapper has a NULL input dataset, which ultimately causes the crash. Now to figure out why that is...
(0032533)
David C. Lonie (developer)
2014-03-24 15:34

Patched volume mappers to handle NULL vtkPoints instances gracefully.
(0032539)
Utkarsh Ayachit (administrator)
2014-03-28 11:55

---------------------------------------------
Topics merged into master:
        14557-color-legend-update
        14580-top-level-checkbox
(VTK) 14615-fix-mpi-segfault
        14650_check_bounds
        catalyst-web
        delaunay3d-alpha-bools
        fix_qt5_dashboard_failures
        pvweb-doc-update
(VTK) vtk_02_20_2014
        14632-find-data-query-combobo
(0032555)
Alan Scott (manager)
2014-04-08 19:07

Nice. Tested remote server, Linux, master.
(0033851)
Erwin (reporter)
2014-11-26 20:30

Issue still persists on a local Windows 7 X64 desktop. Following the steps for Exercise 2.14 of the tutorial book ParaviewTutorial42 I get the following error:

ERROR: In C:\DBD\pvs-x64\paraview\src\paraview\VTK\Rendering\VolumeOpenGL\vtkOpenGLProjectedTetrahedraMapper.cxx, line 185
vtkOpenGLProjectedTetrahedraMapper (00000000168447A0): The required extensions are not supported.


ERROR: In C:\DBD\pvs-x64\paraview\src\paraview\VTK\Rendering\VolumeOpenGL\vtkOpenGLProjectedTetrahedraMapper.cxx, line 185
vtkOpenGLProjectedTetrahedraMapper (000000000B765D40): The required extensions are not supported.

Using Paraview version 4.2.0 64-bit
(0033852)
Alan Scott (manager)
2014-11-26 20:56

Erwin,
There have been a few problems with volume rendering (and Mesa) that have been fixed in the developers tree since ParaView 4.2.0. These fixes will be rolled out in the next few months as ParaView 4.3.0. Further, this bug was for remote server - your bug is local server. Thus, I am closing this bug.

If you think there is a bug in volume rendering, please write up a new bug report. Assuming you are testing local server (Windows, not connecting to a remote server), you could also download a nightly from Kitware, and try that. Further, guessing here - I bet you have older OpenGL drivers that is causing this issue in 4.2.0 (the issue will be masked in 4.3.0). You may want to try updating your graphics drivers.

As always, thanks for reporting bugs! That's one of the wonders of OpenSource - we all get to work together to create a better product. Thank you.

Alan

 Issue History
Date Modified Username Field Change
2014-03-03 15:46 Alan Scott New Issue
2014-03-03 17:23 Utkarsh Ayachit Note Added: 0032448
2014-03-03 17:26 Alan Scott Note Added: 0032449
2014-03-03 17:27 Utkarsh Ayachit Assigned To => David C. Lonie
2014-03-03 17:28 Utkarsh Ayachit Note Added: 0032451
2014-03-05 13:02 David C. Lonie Status backlog => active development
2014-03-05 14:39 David C. Lonie Note Added: 0032457
2014-03-06 15:46 David C. Lonie Note Added: 0032459
2014-03-11 13:20 David C. Lonie Topic Name => 14615-volume-rendering-tet-crash
2014-03-11 13:20 David C. Lonie Status active development => gatekeeper review
2014-03-11 13:20 David C. Lonie Resolution open => fixed
2014-03-20 08:31 Utkarsh Ayachit Fixed in Version => git-master
2014-03-20 08:32 Utkarsh Ayachit Status gatekeeper review => customer review
2014-03-20 08:32 Utkarsh Ayachit Note Added: 0032493
2014-03-20 21:18 Alan Scott Note Added: 0032508
2014-03-20 21:18 Alan Scott Status customer review => todo
2014-03-20 21:20 Alan Scott Note Added: 0032509
2014-03-20 21:49 Alan Scott Relationship added related to 0014581
2014-03-21 12:41 David C. Lonie Note Added: 0032521
2014-03-21 12:42 David C. Lonie Status todo => active development
2014-03-24 15:34 David C. Lonie Topic Name 14615-volume-rendering-tet-crash => 14615-fix-mpi-segfault
2014-03-24 15:34 David C. Lonie Note Added: 0032533
2014-03-24 15:34 David C. Lonie Status active development => gatekeeper review
2014-03-24 15:34 David C. Lonie Fixed in Version git-master => git-next
2014-03-28 11:55 Utkarsh Ayachit Status gatekeeper review => customer review
2014-03-28 11:55 Utkarsh Ayachit Note Added: 0032539
2014-04-08 19:07 Alan Scott Note Added: 0032555
2014-04-08 19:07 Alan Scott Status customer review => closed
2014-05-20 07:29 Utkarsh Ayachit Fixed in Version git-next => git-master
2014-06-17 09:22 Utkarsh Ayachit Target Version => 4.2
2014-08-29 11:00 Utkarsh Ayachit Fixed in Version git-master => 4.2
2014-11-26 20:30 Erwin Note Added: 0033851
2014-11-26 20:30 Erwin Status closed => backlog
2014-11-26 20:30 Erwin Resolution fixed => reopened
2014-11-26 20:56 Alan Scott Note Added: 0033852
2014-11-26 20:56 Alan Scott Status backlog => closed
2014-11-26 20:56 Alan Scott Resolution reopened => fixed


Copyright © 2000 - 2018 MantisBT Team